0:15

Pushkin, we

0:24

will have a test track and trace

0:26

operation that will be world beating.

0:34

That's Boris Johnson, the British Prime

0:36

Minister, speaking to Parliament with a

0:38

typical jingoistic flourish in May

0:40

twenty twenty. The UK,

0:43

he promised, would have a world beating

0:45

contact tracing system within a few

0:47

days. The first wave of

0:49

the pandemic was slowly receding,

0:52

but the cost had been brutal. That

0:55

spring, the country had suffered one

0:57

of the deadliest outbreaks of COVID

1:00

anywhere in the world, and so

1:02

the British Prime Minister decided to

1:04

cheer up the nation as only he could,

1:07

by boasting of a contact trace system

1:09

that would be better than anything Johnny

1:12

Foreigner would have. A

1:15

contact tracing system for COVID has

1:17

three key elements. First,

1:20

you need to be able to identify who's infected

1:23

and isolate them. Then

1:25

you need to trace the recent contacts of

1:27

the infected person. Finally,

1:30

you need to be able to persuade those contacts

1:32

to isolate themselves as well to avoid

1:35

any further spread. It's

1:37

not easy, but if you get

1:39

it right, you can keep the virus

1:41

contained while allowing everyone

1:43

else to relax a little and go about their

1:46

lives. Taiwan managed this,

1:48

so did South Korea and Vietnam

1:51

and Japan. Anyway,

1:53

the UK system didn't need to beat

1:55

the world, just needed to beat the

1:58

virus. In the summer of

2:00

twenty twenty, there seemed to be every

2:02

chance of doing that. Infections

2:05

had been beaten back by a long, strict

2:08

and lockdown. Every

2:11

day there was just a handful of deaths

2:13

and a quarter of a million. Daily tests

2:16

were revealing just a few hundred cases,

2:19

surely few enough for the contact

2:21

tracers to manage. But

2:24

by September there were alarming

2:26

signs that the virus was coming back.

2:29

Cases rose. There were

2:31

a thousand a day, then two

2:33

thousand, then three thousand.

2:36

The testing system was struggling to

2:38

keep up. More than ninety

2:40

percent of people weren't getting test

2:42

results back the next day. There's

2:44

not much point in contact tracing if

2:47

it takes days to even figure out who

2:49

was infected. And then

2:52

on Sunday of the fourth of October,

2:54

my cell phone rang. On

2:56

the line was a researcher from the UK's

2:59

most influential news program.

3:01

She told me that something very odd

3:03

had happened and they wanted me to figure

3:05

it out and then come on the show the next morning

3:07

to help them explain it and

3:10

what exactly had happened. Nearly

3:13

sixteen thousand positive

3:15

cases had disappeared completely

3:17

from the contact tracing system.

3:20

Sixteen thousand people who should have been

3:22

warned that they were infected and a danger

3:24

to others. Sixteen thousand

3:27

cases in which the contact tracers should

3:29

be hurrying to figure out where that infected

3:31

person went, who they met, and who

3:33

else might be at risk. None

3:36

of that was happening. And

3:39

why had the cases disappeared? Well,

3:42

this was the real eye opener. Apparently

3:46

Microsoft Excel had

3:48

run out of numbers. I'm

3:51

Tim Harford, and you're listening

3:54

to cautionary tales. You

4:18

cannot see a crow in a bowl full of

4:20

milk. This

4:22

is Francesco di Marco d'attini.

4:25

He's a textile merchant who lives

4:28

near Florence in Italy. I

4:30

should probably tell you that courtesy

4:33

of Iris Orego's book The

4:35

Merchant of Prato, I've

4:38

taken you back in time. It's

4:40

thirteen ninety six and

4:43

Dattini is furious

4:46

you could lose your weight from your nose to your

4:48

mouth. What was going

4:50

on? Well, D'ttini's

4:52

business associates were bungling

4:54

the numbers. It was a common enough

4:57

problem for any businessman. At

4:59

the end of the thirteen hundreds, Italian

5:02

commerce was becoming complicated. Merchants

5:05

were no longer mere traveling salesman

5:07

able to keep track of profits by patting

5:10

their purses. They were in charge

5:12

of complex operations. The

5:14

Teeney, for example, ordered wool

5:17

from the island of Mayorka two years

5:19

ago before the sheep had even grown.

5:21

It That wool would eventually

5:23

be processed by numerous subcontractors

5:26

before becoming beautiful rolls of dyed

5:29

cloth. The supply chain

5:31

between shepherd and consumer ranged

5:33

across Barcelona, Pisa, Venice,

5:36

Valencia, North Africa, even

5:38

back to Mayorka itself. It would

5:40

be four years between the initial

5:43

order of wool and the final sale

5:45

of cloth. No wonder,

5:48

the Teeney insisted on absolute

5:50

clarity both about where

5:52

the product was at any moment and

5:55

his money. How

5:57

did he manage this simple

6:00

spreadsheets? The

6:03

Teeny, of course, did not use Microsoft

6:06

Excel back in thirteen ninety

6:08

six, but he did use

6:10

its direct predecessor, sheets

6:13

of paper laid out according

6:15

to the system of double entry book

6:17

keeping otherwise known as book

6:19

keeping a la vanetziana.

6:23

In double entry book keeping,

6:25

every entry is made twice the

6:28

clues in the name. For

6:31

example, if you spend a hundred

6:33

florins on wool, that's recorded

6:35

as a credit of a hundred florins in your

6:37

cash account and the debit of

6:40

a hundred florins worth of wool in

6:42

your assets account. This

6:45

extra effort makes it much easier

6:47

to detect if a mistake has been made,

6:50

the books won't balance. Double

6:53

entry bookkeeping became an essential

6:55

method for keeping track of who,

6:57

of what to whom, foreign

7:00

exchange transactions, profits,

7:03

losses, everything. It

7:05

helped merchants such as deteining ensure

7:08

that, no matter how incompetent

7:10

their associates, nothing

7:13

was lost. A century

7:16

later, the master of double entry

7:18

bookkeeping was a man named

7:21

Luca Paccioli. He

7:23

was a serious mathematician, a friend

7:25

of Leonardo da Vinci, but he's

7:27

best known today as the most famous

7:30

accountant who ever lived. He

7:33

literally wrote the book on the double entry

7:35

method back in fourteen ninety

7:37

four. Paccioli once

7:39

advised, if you cannot

7:41

be a good accountant, you will grope

7:44

your way forward like a blind man, and

7:47

may meet great losses. If

7:50

you can't keep your spreadsheets straight,

7:53

you may meet great losses. Remember

7:55

that. Let's

7:58

jump forward nearly five hundred years

8:00

to nineteen seventy eight. We're

8:03

at Harvard Business School and

8:06

a student named Dan Bricklin is

8:08

sitting in classroom watching his

8:10

accounting professor filling in rows and

8:12

columns on the blackboard. The

8:15

professor makes a change and then works

8:17

across and down the grid, erasing

8:19

and rewriting other numbers to make everything

8:22

add up. This

8:24

erasing and rewriting is happening every

8:26

day, millions of times a day,

8:29

all over the world, as accounting

8:31

clerks adjust the entries in what they

8:33

call their spreadsheets, big

8:36

sheets of paper spread across two

8:38

pages of an accounting ledger. Adjustments

8:41

take a lot of work. But

8:44

Dan Bricklin was a computer geek, a

8:47

former programmer, who immediately

8:49

thought, I can do this on

8:51

a computer. You would put a number

8:53

in and hit return, and everything would recalculate,

8:56

and you could watch it. You could watch the number

8:58

change. Bomb bomb bomb, It made a sound.

9:01

I had a real prototype. The

9:07

rest is history. Bricklin

9:11

and a friend called their spreadsheet

9:13

program VisiCalc. It

9:15

went on sale on the seventeenth of October

9:18

nineteen seventy nine. It was a smash

9:20

hit, soon followed by Lotus

9:23

one, two three, and then

9:25

in due course by Microsoft

9:27

Excel itself. For

9:30

accountance, digital spreadsheets

9:32

were revolutionary, replacing

9:34

hours of painstaking work with

9:37

a tap on a keyboard. But

9:39

some things did not change. Accountants

9:43

still had their professional training and

9:45

their double entry system.

9:48

But for the rest of us, well,

9:50

Excel became ubiquitous, an

9:53

easily accessible tool, a flexible

9:55

tool like a Swiss Army penknife,

9:58

sitting in your digital back pocket. Any

10:01

idiot could use it, and we

10:03

did. Oh goodness me, we

10:06

did. Nobody

10:13

really knows what happened to the fifteen

10:15

thousand, eight hundred and forty one

10:18

positive COVID cases that disappeared

10:20

from the spreadsheet. Public Health

10:23

England, a government agency responsible

10:25

for the process, hasn't published

10:27

anything very informative on the issue. I

10:30

asked them about it. The suggestion

10:32

that any cases were lost is simply

10:35

incorrect. Oh, no

10:37

cases were missed. There was a delay in

10:39

referring cases for contact tracing

10:41

and reporting them in the national figures.

10:44

That delay was often four or five

10:47

days. But experts will

10:49

tell you that you really need to track contacts

10:52

within forty eight hours. A five

10:54

day delay renders the test results

10:57

almost useless. Look,

10:59

guys, if you lose the positive cases

11:02

for four or five days, you

11:04

lose them. But how

11:07

did they lose them?

11:09

Somewhere? Somehow? Public

11:12

Health England had used the wrong Excel

11:15

file format XLS

11:18

rather than the more recent XLSX

11:21

and XLS spreadsheets simply

11:24

don't have that many rows to

11:26

to the power of sixteen about sixty

11:28

four thousand. That meant that during

11:31

some automated process cases

11:34

had vanished off the bottom of the spreadsheet

11:36

and nobody had noticed. The

11:40

idea of simply running out of space

11:42

to put the numbers was rather amusing. The

11:45

Fact that Microsoft was never anyone's

11:47

idea of cool simply added

11:49

to the hilarity. Do you suffer

11:52

from having to organize and analyze a small

11:54

set of numbers? Is the

11:56

undue function on a calculator frustrating

11:58

the underpowered for your calculations needs?

12:01

Do you want to dabble in recreational mathematics,

12:04

then spreadsheets maybe for you.

12:07

That's a satirical advert from the

12:10

comedy and mathematics YouTube channel

12:12

Stand Up Maths. Please speak

12:14

to your database developer before deciding if spreadsheets

12:17

is right for you. Common side effects include

12:19

accidentally sorting some but not all of

12:21

the data, slight cell loss when selecting

12:24

numbers, hashtag name, question

12:26

mark, losing key medical

12:28

data during a pandemic and endangering lives,

12:31

and being fired. Spreadsheet

12:33

is intended for short term use only. Stop

12:35

using spreadsheets if you find yourself in charge of a government

12:38

database with life and death ramifications. Spreadsheets

12:41

from the makers of word art. A

12:44

few weeks after the data loss scandal,

12:47

by a strange twist of fate, I

12:49

found myself able to ask Bill

12:51

Gates himself about what had

12:53

happened. Bill

12:56

Gates no longer runs Microsoft, and

12:58

I was interviewing him about vaccines for

13:00

a BBC program called How to

13:02

Vaccinate the World, but the opportunity

13:05

to have a bit of fun quizzing him about XLS

13:07

and xlsx it's too good to miss.

13:10

I expressed the question in the nerdiest

13:13

way possible, and Gates's

13:15

response was so straight laced I

13:17

had to smile to myself. Yeah,

13:20

I guess the older format. You

13:22

know, they overran the sixty four thousand

13:24

limit, which is not there in the new

13:26

format, So you

13:30

know, it's good to have people double check

13:32

things, and I, you know,

13:34

I'm sorry that happened. Exactly

13:37

how the outdated XLS format

13:39

came to be used is unclear. Public

13:42

Health England sent me an explanation, but it

13:44

was rather vague. I didn't

13:46

understand it, so I showed it to some members

13:48

of use Brig, the European

13:51

Spreadsheet Risks Group. They spend

13:53

their lives analyzing what happens when

13:55

spreadsheets go rogue. They're my

13:57

kind of people. But they

14:00

didn't understand what Public Health England had

14:02

told me either. It was all

14:04

a little light on detail. The

14:07

basic problem was that whatever like

14:09

Health England had done wrong, they

14:11

didn't have the right checks and controls

14:14

to flag up problems. But

14:16

I can just imagine what the merchant of Prato,

14:19

Francesco DiMarco D'ttini, might have

14:21

said. You could lose your way from

14:23

your nose to your mouth. We'll

14:28

explore how Excel became

14:30

so error prone after this

14:33

message. Doctor

14:38

Felina Herman's is a researcher who

14:41

studies spreadsheets. A

14:43

few years ago, she realized that there was a

14:45

wonderful source of spreadsheets that she

14:47

could study in their natural habitat.

14:50

That source was a bankrupt energy

14:52

company called Enron. Enron

14:55

used to be huge, but two decades

14:58

ago it collapsed and various

15:00

Enron executives were convicted of financial

15:03

crimes. Regulators extracted

15:05

a large digital pile of half

15:07

a million emails from and run servers,

15:10

and those emails are publicly

15:12

available. Importantly,

15:14

for doctor Herman's, thousands

15:16

of those emails had spreadsheets

15:19

attached. She started

15:21

digging through them. Looking at nearly

15:24

ten thousand spreadsheets with calculations

15:26

in them, she found that a quarter of them

15:29

had at least one obvious

15:31

error. The errors even

15:33

seemed to multiply. If a spreadsheet

15:36

had any mistakes at all, on

15:38

average, it contained more than seven

15:40

hundred and fifty

15:43

How can a spreadsheet acquire so many

15:45

errors? I asked my friend Matt

15:48

Parker, the man who literally wrote

15:50

the book about mathematical mishaps and their

15:52

consequences, a book with a

15:54

delightful title Humble Pie

15:57

Imagine Cautionary Tales, only

15:59

with more jokes and more equations. One

16:02

spreadsheet problem is simple human

16:05

error. For example, the time

16:07

when candidates for a job in policing

16:10

were listed alongside a column containing

16:12

their scores on a test. When

16:14

one column was resorted and

16:17

the adjacent one was not, the

16:19

test scores were effectively

16:21

scrambled all the time

16:23

that the investment bank JP Morgan

16:26

lost six billion dollars.

16:28

And when I say lost, I

16:30

mean they lost the money, not that they misplaced

16:32

it for five days. They

16:35

lost this six billion dollars after

16:37

several spreadsheet errors, notably

16:39

one in which a risk indicator in a spreadsheet

16:42

was being divided not by an average

16:44

of two numbers but by their sum.

16:47

That made the risks look half as big as

16:49

they should have done. But

16:51

Excel is happy to introduce errors

16:53

without any help from US humans. Matt

16:57

Parker told me that one common set of

16:59

problems is produced by the auto

17:01

correct function. Excel

17:04

loves to autocorrect. Type

17:06

in an international phone number, and

17:09

Excel will strip off the leading zeros.

17:11

They're mathematically redundant, but if

17:14

you want to make a phone call, you'll find

17:16

that they're not redundant at all. Or

17:19

if instead you type in a twenty digit

17:22

serial number, Xcel will

17:24

decide those twenty digits are

17:26

a huge quantity and round them off,

17:29

turning the last few digits

17:31

into zeros. If

17:33

you're a genetics researcher typing in

17:35

the name of a gene such as march

17:37

f one or sept in one are

17:39

generally abbreviated to march

17:42

one or sept one. Well,

17:45

you can imagine what Xcel does with them.

17:48

It turns those gene names into dates,

17:52

and one study estimated that

17:54

twenty percent of all genetics

17:57

papers had errors caused

17:59

by Xcel's autocorrection. Microsoft's

18:04

response to the genes problem is

18:06

that Xcel's default settings

18:08

are intended to work in most day to day scenarios,

18:10

which is the polite way of saying, guys,

18:14

Excel was designed for accountants, not

18:16

genetics researchers. But

18:19

it's understandable that scientists picked

18:21

up Excel and started to use it.

18:23

It's right there on every computer. It's

18:26

powerful, it's flexible, it's

18:28

ubiquitous. The

18:30

problem with ubiquitous tools is that

18:32

we tend to use them even when they

18:34

aren't the right tool for the job, even

18:37

when we don't really know what we're doing. Come

18:39

to think of it, especially

18:42

when we don't know what we're doing. I

18:44

said earlier that Microsoft Excel

18:46

is like a Swiss army knife. As

18:49

a boy, I was absolutely fascinated

18:51

by these beautiful little red multi

18:54

tools, a pen knife with a

18:56

can opener and three kinds of screwdriver,

18:58

and a bottle opener, and a wire stripper and a tiny

19:01

saw and some tweezers and even a toothpick.

19:03

What a world of miracles and wonders.

19:07

But as an adult Gig struggles

19:09

to put up a bookshelf straight even.

19:11

I've noticed something about people

19:14

with practical skills, people

19:16

such as plumbers, electricians,

19:18

and carpenters. They don't

19:21

use a Swiss army knife. They

19:23

bring a toolkit with professional

19:25

tools. Microsoft

19:27

Excel is a professional enough tool

19:30

if you're an accountant. Excel

19:33

wasn't designed to run the entire contact

19:35

tracing infrastructure of a wants proud

19:37

nation any more than a Swiss army knife

19:40

was designed to help you put up a set of shelves.

19:43

The experts I've spoken to have different

19:45

views about the deeper problem here. Some

19:48

of them reckon that using Excel itself

19:50

for contact tracing was the original

19:52

sin, that a different sort of software

19:55

tool, a database, would have been much

19:57

more appropriate. Others say

19:59

no, if you use Excel professionally

20:02

with proper controls, it can easily

20:04

handle the task of contact tracing. And

20:07

a well designed database would have taken time

20:09

to implement. XCEL was right there.

20:12

Professional carpenters don't use a Swiss

20:15

army knife. But if the shelves need to

20:17

be put up immediately and you don't have a toolbox,

20:20

why not give the Swiss army knife a try.

20:22

You just have to be aware of its limitations

20:25

and perhaps to redo the job

20:27

properly when you have the tools to do so.

20:33

Not long ago, I asked folks on

20:35

Twitter if they could recommend some good

20:37

books about the eradication of

20:39

smallpox. Most people instead

20:42

recommended books about Edward Jenna

20:45

back in seventeen ninety six, when

20:47

he first demonstrated an effective

20:49

smallpox vaccine. That's

20:52

revealing because I'd asked

20:54

about the eradication of smallpox,

20:56

and smallpox wasn't eradicated in seventeen

20:59

ninety six, not even close.

21:02

And while eradication would have been impossible

21:05

without a highly effective vaccine,

21:07

it also required highly effective

21:10

use of information, or, as

21:12

the merchant Francesco di Marco d'atini

21:15

might have said, it required not

21:17

losing your way from your nose

21:19

to your mouth. Unlike

21:22

COVID, smallpox infections

21:24

are easy to detect. For the awful

21:27

reason that smallpox does so much

21:29

damage to the human body. Bill

21:31

Fagy, one of the leaders of the fight against

21:33

smallpox, says that you can even

21:35

follow your nose. On at

21:38

least two occasions, smell

21:40

alone alerted me to the presence

21:42

of small pox. As I walked

21:44

down a hospital hallway in India,

21:47

the dead animal odor stopped

21:49

me in my tracks. Following

21:52

the smell, I located a

21:54

smallpox patient. Another

21:57

time, as I walked down an alley

21:59

in an urban slim in Pakistan, the

22:01

same smell hit me. There

22:04

are competing smells in such places,

22:07

but again one smells stood

22:09

out. Knocking on

22:12

doors, I found two siblings

22:15

with smallpox. Ever

22:19

since the vaccine for smallpox was

22:21

demonstrated in seventeen ninety six,

22:24

people dreamed of eradicating the disease,

22:27

but those dreams kept failing to

22:29

come true. The vaccinators

22:32

would never manage to reach quite enough

22:34

people in poorer countries,

22:36

smallpox would linger in isolated

22:39

rural communities or neglected slums.

22:41

A generation of babies would be borne

22:44

without any immunity, and soon

22:46

enough the disease would be back. In

22:49

the mid nineteen sixties, smallpox

22:52

was still killing two million

22:54

people a year. This was the

22:57

same number as died of COVID. In twenty

22:59

twenty, the World Health

23:01

Organization announced that it would

23:04

redouble its efforts to eradicate the disease,

23:06

and it planned to do so by intensifying

23:09

the mass vaccination campaign. Bill

23:11

Fegi was part of those efforts

23:13

to fight smallpox. Fegi

23:16

would show up in a village in eastern Nigeria,

23:18

all six foot seven of him, and the

23:20

local elders would put out the word come

23:23

and see the tallest man in

23:25

the world, and people would come,

23:28

and Bill Fegy reckons he wants vaccinated.

23:31

Eleven thousand, six hundred people

23:34

in a single day. It

23:36

wasn't enough. Still, The

23:38

outbreaks came late

23:41

in nineteen sixty six. Vegi

23:43

received a radio message. This

23:46

is a message for doctor Fagi, A

23:48

message for doctor Segy Veggie

23:51

speaking what is it? We'll

23:54

hear that message and why

23:57

information matters if you want to eradicate

23:59

smallpox. After the break, This

24:10

is a message for doctor Pegi. A

24:13

message for doctor Segy Veggie

24:16

speaking what is it? The

24:18

radio operator told doctor Bill Fegi

24:21

that there had been an outbreak of smallpox

24:23

in a village about one hundred miles away. He

24:26

traveled there, found five cases and

24:28

vaccinated everyone they'd been in contact

24:30

with. The handy thing about the smallpox

24:33

vaccine is that it often still

24:35

works even if you vaccinate someone

24:37

a few days after they've been exposed

24:40

to the virus. Standard

24:42

practice then would be to vaccinate everyone

24:44

for miles around, but Vega's

24:46

team just didn't have enough doses

24:48

with them, so instead he used

24:50

radio and the local network of missionaries

24:53

to try to work out where to use the

24:55

vaccine. Every evening at

24:57

seven o'clock they'd switch on the radio

25:00

and put the word out, this

25:03

is doctor Bill Fegy speaking here.

25:06

Doctor Pegi, which send out

25:11

and we have all the information

25:13

you requested. That's amazing

25:16

news. So are there any new cases?

25:21

Cases were identified in just four

25:23

more villages. Vega and his

25:26

team quickly raced to the scene and administered

25:28

the vaccine. The hope was that

25:30

the vaccines would act like a firebreak

25:33

the disease wouldn't find anyone to spread

25:35

to, and it worked. Repeating

25:38

the tactic, Vega's team eliminated

25:41

smallpox from eastern Nigeria

25:44

within six months, just in

25:46

time for the catastrophic civil war of

25:48

nineteen sixty seven. Despite

25:51

the chaos and enormous bloodshed of

25:53

that war, smallpox did

25:55

not return. The

25:58

secret to the success was to worry less

26:00

about the blanket coverage that was never

26:03

quite good enough, and worry more

26:05

about quickly finding exactly

26:07

where each outbreak was. Eradication

26:11

was all about information, and up

26:13

until that point information had

26:16

been very patchy. As the WHO

26:19

teams looked more closely, they

26:21

realized they were missing the vast majority

26:23

of the cases. Instead of one

26:25

hundred thousand cases a year around the

26:27

world, there were ten million.

26:31

Public health workers could beat smallpox

26:33

by figuring out quickly where the outbreaks

26:36

were and swiftly controlling the situation,

26:39

isolating people with the disease and vaccinating

26:42

their contacts. The strategy

26:45

became known as a ring vaccination,

26:48

and it has a lot in common with COVID

26:50

contact tracing. In both

26:52

cases, you need to rapidly isolate

26:54

infected people and find their recent

26:57

contacts. Ring vaccination

26:59

worked, and it didn't take

27:01

long. The last

27:04

gasp of smallpox in the wild was

27:06

in Somalia late in

27:09

nineteen seventy seven. Ali

27:11

Mayaw Marlin, twenty three years

27:13

old, a cook and part time

27:16

vaccinator, had astonishingly

27:19

not been vaccinated himself. One

27:22

day, he was asked for directions to

27:24

the local hospital by a man driving a

27:26

jeep with two sick children in the

27:28

back. Soon enough, he

27:31

started to feel unwell. He

27:33

was wrongly diagnosed first with malaria

27:36

and then with chicken pox. He

27:38

wasn't isolated or treated until

27:40

a friend of his, a nurse, made the correct

27:43

diagnosis. Ali had

27:45

the awful smallpox. His

27:48

ninety one friends and contacts

27:50

were isolated and vaccinated.

27:54

None of them contracted the disease. Ali

27:57

himself recovered and devoted

28:00

his life to the fight against

28:02

polio. I tell them how important

28:04

these vaccines are. I tell them not to do

28:07

something foolish like me. And

28:10

the vaccines were important, essential

28:12

in fact, but so was

28:15

quickly identifying and tracing contacts

28:18

at risk. Smallpox

28:20

had survived nearly two centuries

28:23

of vaccination, but

28:25

it couldn't survive a well run system

28:28

that targeted outbreaks and tracked

28:30

potential cases with hindsight.

28:33

It seems so easy and simple in

28:36

a way it was, But of course

28:38

keeping track of things is harder than it

28:41

might first appear. Francesco

28:43

di Marco D'ttini could have told you that so

28:46

could Bill Gates. If

28:51

you really want proof that contact tracing

28:53

works, how would you get it? If

28:56

you were a mad scientist, praised

28:58

with power and unchained by conventional

29:01

ethics, You'd do an experiment.

29:04

You'd hack into a country's contact tracing

29:06

system. Then you'd delete some of the positive

29:08

case, making sure that some

29:11

regions lost a lot of cases and

29:13

some lost very few. Then

29:15

you'd compare what happened in the places

29:17

where the contact tracing system was still running

29:20

smoothly to the places where thousands

29:22

of cases had gone missing. If

29:25

you weren't an evil genius, of course, you

29:28

wouldn't dream of doing such a thing. Instead,

29:31

you'd keep an eye out for it happening by accident

29:34

because somebody bungled the formatting

29:36

of Excel spreadsheets. Two

29:40

economists, Timo Fetzer

29:42

and Thomas Graber did just that.

29:46

They decided that no catastrophe

29:48

should be allowed to occur without

29:50

trying to learn some lessons, which is very

29:53

much in the spirit of cautionary tales.

29:55

They combed through the evidence from

29:58

Public Health England's mishap, and

30:00

by comparing the different experiences

30:02

of different regions, they concluded

30:05

that the error had led to one hundred

30:07

and twenty five thousand additional

30:09

infections. The story

30:11

about Excel running out of numbers

30:14

just seemed so funny at first. Do

30:17

you suffer from having to organize and analyze

30:19

a small set of numbers? And

30:22

Bill Gates's straight faced, straight

30:24

laced response seemed funny too. Yeah,

30:27

I guess the older format. You

30:29

know, they overran the sixty four thousand

30:31

limit, which is not there in the

30:33

new format. You know, it's

30:36

good to have people double check things, and you

30:38

know, I'm sorry that happened. But

30:40

of course it was Gates who'd

30:43

seen through the joke on the surface

30:46

to what lay beneath. He wasn't

30:48

laughing, not because he had no sense of humor,

30:51

but because he understood that this wasn't a

30:53

comedy. It was a tragedy.

30:57

The economists Fetza and Graber

30:59

have calculated a conservative estimate

31:02

of the number of people who died unknown

31:05

victims of the spreadsheet error.

31:08

They think the death toll is at least fifteen

31:11

hundred people. So

31:15

the next time there's a pandemic, let's

31:17

make sure we have our spreadsheets in order.

31:21

After all, As Leonardo da Vinci's

31:23

friend, the father of accounting, Luca

31:25

Paccioli warned us more than

31:28

five hundred years ago, if

31:30

you cannot be a good accountant, you

31:33

will grope your way forward like a blind

31:35

man and may meet great

31:38

losses. Fifteen

31:40

hundred people dead, great

31:42

losses indeed. Key

32:02

sources for this cautionary tale

32:04

include Planet Money episode six

32:07

h six, Spreadsheets, Matt

32:10

Parker's YouTube video whence

32:12

Spreadsheets Attack, and Bill

32:14

Feggy's book House on Fire.

32:17

For a full list of our sources, see

32:19

Tim Harford dot com. Cautionary

32:25

Tales is written by me Tim

32:27

Harford with Andrew Wright. It's

32:29

produced by Ryan Dilley and Marilyn

32:31

Rust. The sound design and original

32:34

music are the work of Pascal Wise.

32:37

Julia Barton edited the scripts.

32:40

Starring in this series of Cautionary

32:42

Tales are Helena Bonham, Carter

32:44

and Jeoffrey Wright, alongside

32:47

Nazar Alderazzi, Ed Gochen,

32:50

Melanie Gutteridge, Rachel Hanshaw,

32:53

cobnor Holbrook, Smith, Reg

32:55

Lockett, Missiamunroe and

32:57

Rufus Wright. The show would

32:59

not have been possible without the work of Mia

33:02

LaBelle, Jacob Weisberg, Hella

33:04

Fane, John Schnarz, Carlie

33:07

mcgliori, Eric Sandler, Emily

33:09

Rostock, Maggie Taylor, Daniella

33:12

Lakhan, and Maya Kane. Cautionary

33:15

Tales is a production of Pushkin

33:17

Industries. If you like the show, please

33:20

remember to share, rate, and

33:22

review.